CentreForDigitalHumanities / CentreForDigitalHumanities/programming-in-python
Python is a bit under-equipped for statistical modeling
- Dominant language
- Jupyter Notebook
- Stars
- 1
- Forks
- 1
- PR merge metrics
- No merged PRs in 30d
Description
One of the course participants was trying to fit a linear mixed effects model with two levels of grouping. [`statsmodels.MixedLM`](https://www.statsmodels.org/stable/generated/statsmodels.regression.mixed_linear_model.MixedLM.html#statsmodels.regression.mixed_linear_model.MixedLM) does not support this. You can interface with R packages that provide this functionality through [pymer4](https://eshinjolly.com/pymer4/), but installing that in a Jupyter notebook is nontrivial because it also involves installing the dependent R packages.
Take home point: you can do statistical modeling in Python, but if your use case is somewhat advanced, you are probably better off using something more special-purpose such as R. We should add this as a note to the statistical modeling section in the tips.
Contributor guide
Research direction
Locate the statistical modeling section in the tips and read its surrounding guidance on Python modeling. Add a note explaining the limitation of statsmodels.MixedLM for two grouping levels and the suitability of more specialized tools such as R, then verify that the note fits the section and accurately reflects the linked resources.
Written by the indexing model from the issue text.
Assessment
- Tech stack
- python, r
- Domain
- documentation
- Issue type
- Documentation
- Difficulty
- 2/5
- Estimated time
- 1-3 hours
- Activity status
- Stale
- Clarity
- Mostly clear
- Newbie friendliness
- 50/100